MCP Integration Expert
Expert in Model Context Protocol (MCP) for connecting AI applications to external systems, data sources, and tools. Specializes in building MCP servers and clients following the official specification.
What This Skill Does
This skill provides expert guidance on implementing the Model Context Protocol (MCP), an open-source standard for connecting AI applications to external systems. MCP enables AI applications like Claude or ChatGPT to connect to data sources (local files, databases), tools (search engines, calculators), and workflows (specialized prompts).
Think of MCP like a USB-C port for AI applications—it provides a standardized way to connect AI applications to external systems.
Instructions
When working with MCP integration tasks, follow these steps:
1. Understand the Use Case
Determine if the user needs to **build an MCP server** (expose data/tools) or **build an MCP client** (connect to MCP servers)Identify what external systems, data sources, or tools need to be connectedClarify the AI application or agent that will use the MCP integration2. MCP Server Development
When building an MCP server to expose data and tools:
Design the server architecture following MCP specificationsImplement standardized endpoints for: - **Resources**: Data sources the AI can access (files, databases, APIs)
- **Tools**: Actions the AI can perform (search, calculate, execute)
- **Prompts**: Specialized prompts or workflows
Ensure proper authentication and security measuresImplement versioning support for compatibilityAdd comprehensive error handling and loggingDocument available resources, tools, and prompts clearly3. MCP Client Development
When building an MCP client application:
Implement MCP protocol communication following the specificationHandle connection management to MCP servers (local and remote)Implement discovery mechanisms for available resources and toolsBuild request/response handling for MCP operationsAdd proper error handling and retry logicImplement caching strategies where appropriateEnsure compatibility with MCP versioning4. Local vs Remote Connections
**For local MCP servers:**
Set up local development environmentConfigure file system or local database accessImplement inter-process communicationTest with local AI applications**For remote MCP servers:**
Implement network communication (HTTP/HTTPS, WebSocket)Add authentication mechanisms (API keys, OAuth)Handle network errors and timeoutsImplement rate limiting and quota management5. Security Best Practices
Never expose sensitive data without proper authenticationImplement rate limiting to prevent abuseValidate all inputs from AI applicationsUse encryption for data in transitFollow least-privilege principles for tool accessImplement audit logging for all operationsAdd security headers and CORS policies6. Testing and Validation
Use the MCP Inspector tool for debuggingTest with multiple AI applications/clientsValidate protocol complianceTest error scenarios and edge casesVerify performance under loadEnsure backward compatibility7. Documentation
Provide clear documentation including:
Available resources, tools, and promptsAuthentication requirementsRate limits and quotasExample requests and responsesError codes and handlingVersioning informationExample Use Cases
1. **Personal AI Assistant**: Connect AI to Google Calendar and Notion for personalized assistance
2. **Code Generation**: Enable Claude Code to use Figma designs for web app generation
3. **Enterprise Chatbots**: Connect to multiple databases for organization-wide data analysis
4. **Creative Workflows**: Connect AI to Blender for 3D design and 3D printer integration
5. **Development Tools**: Integrate code repositories, testing frameworks, and deployment systems
Key Concepts
**Servers**: Expose data sources and tools through standardized MCP endpoints**Clients**: AI applications that connect to and consume MCP servers**Resources**: Data that can be accessed (files, database records, API responses)**Tools**: Actions that can be performed (search, calculate, execute commands)**Prompts**: Specialized prompt templates or workflows**Versioning**: Protocol version negotiation for compatibilityImportant Notes
Always refer to the official MCP specification at https://modelcontextprotocol.io for the latest standardsMCP is open-source and designed for interoperability across AI applicationsFocus on standardization—MCP's value comes from consistent implementationConsider using existing MCP SDKs to accelerate developmentSecurity and privacy are critical when exposing data to AI applicationsTest thoroughly with the MCP Inspector before production deploymentResources
Official Documentation: https://modelcontextprotocol.ioSpecification: Available at modelcontextprotocol.io/specificationMCP Inspector: Developer tool for debugging MCP implementationsGitHub: Check the official MCP GitHub repository for examples and SDKs